This direction studies cities as open spatial environments and builds intelligence that can perceive, model, reason about, and act within them. It connects multimodal spatial reasoning across remote sensing, street views, and 3D scenes; geospatial intelligence over structured urban data; world models for simulating open environments; and physical AI for spatial action and real-world applications.
This direction studies agents that can reason, coordinate, and interact with people, tools, and environments. It builds embodied agents, socially grounded agents, and multi-agent collaboration mechanisms, with emphasis on reinforced reasoning, multimodal agent interaction, social simulation, human-AI hybrid coordination, and deployment in governance and industry-facing applications.
This direction focuses on urban analytics and data-grounded modeling of human behavior in cities, including activity understanding, mobility patterns, social sensing, population dynamics, and governance-related urban phenomena. It emphasizes interpretable evidence and deployable insights that connect large-scale urban data with time-series modeling, urban governance, planning, policy evaluation, and public-service applications.
A systematic summary of my earlier work in individual mobility behavior, citywide spatio-temporal prediction, and operational urban decision systems.